Neural networks basics
Artificial Intelligence · Engineering
Study notes
Learn OR gate: perceptron with weights w1,w2, bias b; output 1 if w1*x1+w2*x2+b > 0. Start random; for each example, if wrong, adjust weights toward correct. After epochs, it learns OR. Single perceptrons can't learn XOR (not linearly separable); hidden layers fix this: the birth of deep learning.